Why Normal People Aren’t Using AI Agents
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Josh Miller argued that AI agents are not a product but a technical frame, and consumers don’t care about the underlying technology as long as it solves their problems
- OpenAI reported its Codex and ChatGPT Work agents have around 10 million weekly users, a fraction of the billion monthly users its chatbots attract
- Anthropic's Claude Code and Cowork agents are seeing adoption levels similar to OpenAI’s, but still represent a tiny share of overall generative AI usage
- Miller said no one outside tech has mentioned using an AI agent, underscoring a disconnect between industry enthusiasm and public indifference
- The Browser Company built a popular AI-powered morning briefing in its browser Dia, which technically uses an agent but hides the complexity from users
- Miller criticized AI labs for fixating on sci-fi visions like the movie Her, leading to groupthink and a lack of diverse product thinking
Why it matters: AI labs have invested billions in agent-capable models, but with only 10 million weekly users across OpenAI and Anthropic versus nearly a billion for chatbots, the market shows these tools remain niche. Without shifting from tech-driven demos to user-centered design, the promised transformation of work and life won’t materialize.
